Replacing a Plc-based control to a computer control

L

Leonard Misner

"PC with a simple robust OS" is an oxymoron.
there is no blue screen of death on a PLC.

You can talk about how fast your PC is and how it runs circles around that x86 PLC...until you have your PC running soft PLC, driving animated graphics, handling a hundred analog process devices, calculating statistics and creating paretos and productions reports...and when you can't even get 4 temperature samples while attempting to monitor a single analog device at a tenth-of-a-second sample rate, then tell me how a pc is any comparison at all to a plc. Consider: Why does a PLC exist at all? Why was it created? If a PC architecture is EVER the optimal option, how does 'plc' ever exist and continue to exist? Do you have some magic PC that doesn't slow down whenever you open more and more applications?

A pc is a single tasking slow stupid device which is incapable of having a few hundred process applications running simultaneously while still processing the entire program, scanning all I/O and updating the image table repeatedly and absolutely reliably at a rate of a few milliseconds even with the plc sitting alone in an oil saturated box for decades at a time.

Yes...there are industrial pc's...but they're still limited by their architecture. I'm speaking from 10 years experience leading multimillion dollar plc based automated assembly and test systems internationally from concept through 6-sigma runoff...add another 25 years of working with automated systems in automotive, pharm and food manufacturing facilities - both plc and pc based systems. Test engineers cry whenever I point out that even their test stands (typically pc based controls) can be done easier and more reliably with a plc. One of the biggest reasons the United States has had so many manufacturing plants closed down is because we simply have lost the ability to being competitive in the world market. You have to get the best quality product out the door at the lowest cost. that's a pretty simple concept, but it seems so many engineers believe the company they are working for is there to facilitate their own personal technical experimentation and entertainment. Not nearly the challenge of a PLC application then to do all that forced integration and special configurations and code of most pc systems. You have mentioned how a PLC is more readily supported, more easily programmed by more people who have that knowledge, ethernet integration is simply a module and copying some config junk from a manual as you suggested. So why would intentionally compromise the future of the company you're working with by forceably integrating a pc with various uniquely concepted techy toys that were NOT DESIGNED SPECIFICALLY TO the task of reliable machine control. If you (or the support team left in your dust) can access your controls, understand the process as programmed and have an ability to improve analog monitoring and controls...then you can improve product quality and production efficiency. The market wants to buy less expensive golden nuggets...not some terd product that was produced by a "black box" that nobody can work with to make that terd product any better than the terd product that it is. If your stuff can't compete in the market, your company shuts down...welcome to America. Engineers are supposed to be a pretty sharp group...why haven't they figured THAT out?

Anybody ever hear of Dr. Deming? ...don't get me started. lol.
 
J

James Ingraham

I know this is going to seem odd, but I'm going to have to disagree with Leonard Misner. Yeah, I know I just spent like 8 billion words arguing with Armin about it, but that was over a fairly nuanced distinction between "PC" and "PLC." I believe Mr. Misner's strong dislike of PCs for control is a bit unwarranted.

Leonard Misner: <i>"PC with a simple robust OS" is an oxymoron.</i>

There is ample evidence to refute this. Many PCs have uptimes comparable to PLCs. QNX, VxWorks, and other RTOSes that run on x86 can run indefinitely. Then there's all those LAMP web servers out there. There are numerous PC-based solutions for control from Beckhoff, B&R, etc. Kuka and ABB robot controllers are PC-based.

Leonard Misner: <i>There is no blue screen of death on a PLC.</i>

An over-rated concern, and not necessarily accurate. PLCs do die, even if it's relatively rare. Most Windows blue screens are the result of hardware problems and/or improperly written drivers. Within the limited environment of a PC-based controller those issues go away, particularly if using a boxed solution (e.g. Beckhoff again). For anecdotal evidence, I have quite a few PCs doing control that have run for a decade or more. Hardware failures is in fact one reason my company moved to PLCs instead of PCs, but even then we're talking about five year or so life spans. Sure, you expect 20 out of your PLC. Of course, I'll spend less over the course of 20 years on hardware even if I have to replace it every four or five years.

And don't forget Stuxnet; it turns out that the only reason PLCs haven't been infected by viruses before was that no one was trying. Security researchers are now looking at the PLC manufacturers and are horrified at the complete lack of security consciousness.

Leonard Misner: <i>You can talk about how fast your PC is and how it runs circles around that x86 PLC...until you have your PC running soft PLC, driving animated graphics, handling a hundred analog process devices, calculating statistics and creating paretos and productions reports...and when you can't even get 4 temperature samples while attempting to monitor a single analog device at a tenth-of-a-second sample rate, then tell me how a pc is any comparison at all to a plc.</i>

It is certainly true that PCs have a lot of overhead that PLCs don't, and that PLCs can take advantage of being closer to the "metal." Nevertheless, a PC can bring some horsepower to bear that PLCs cannot conceivably match. A 3GHz quad-core i7 is on the order of 100 GFLOPS. I can write some pretty crappy software and still run unbelievably fast. The hard part is connecting to real world I/O. Most PLC manufacturer's won't tell you the speed of their backplanes, but I seriously doubt that it's even close to the 500MB/s and 5GT/s of a PCI Express 2.1 x1 slot. And if you REALLY need it, you can jump up to a PCI-E 3.0 x16 for 16GB/s and 128GT/s. Let's round that to a byte per 100 picoseconds. Assuming those 4 temperature sensors need 2 bytes each, I should be able to retrieve them in under a nanosecond. I can run 100 mathematical operations in another nanosecond, then send them back. This is a thought experiment, of course. I doubt you'd actually be able to this in under a microsecond, possibly even tens of microseconds. But I won't have any trouble hitting your 100 milliseconds. Heck, I was doing better than that back when all the PCs we shipped had 486 processors.

Leonard Misner: <i>Consider: Why does a PLC exist at all?</i>

There are lots of reasons, of course. For one thing, when the PLC was invented circa 1970 there was no such thing as a PC. It definitely took some time for the general purpose nature of the PC to get the horsepower needed to compete with a dedicated-purpose box. Even relatively recently, PLCs weren't required to communicate all that much. As businesses have tried more and more to increase productivity, reduce downtime, and match production to demand, it has become more important to feed data from control systems up into management systems. Without that necessity, the PC didn't have as compelling an argument on the shop floor.

Leonard Misner: <i>Why was it created? If a PC architecture is EVER the optimal option, how does 'plc' ever exist and continue to exist?</i>

This is a somewhat specious argument, and can be equally flipped around. If PLCs are ever the best solution, why do PC-based solutions exist and continue to sell well?

Leonard Misner: <i>Do you have some magic PC that doesn't slow down whenever you open more and more applications?</i>

Actually, yes. Most PC-based solutions have some way of dealing with this. Kuka and Beckhoff both run their control software on a RTOS. Windows then runs parallel to (or as a guest of, depending on things are set up) the RTOS. You can launch Crysis and your control will still run just fine. This isn't really necessary, however. I have five robot systems in the real world running on stock Windows 2000. The fact that they are running on Windows 2000 gives you an idea of the age of the machines. It's true that if you launch Crysis on one of THOSE boxes, your robot will not run very well. DON'T DO THAT. We've never had a problem with this. And I certainly allow operators to pull up PDFs etc. on the controller, and it handles that just fine.

Leonard Misner: <i>A pc is a single tasking slow stupid device</i>

Is this deliberate trolling? Single tasking? Slow? This hasn't been true for at least 15 years.

Leonard Misner: <i>which is incapable of having a few hundred process applications running simultaneously</i>

Some high-end servers (x86 or otherwise) can probably handle a few hundred process applications simultaneously. Obviously, this sounds like a clustering or mainframe application. Certainly not something you could do on a PLC. Most PLCs can't handle a few hundred PID loops, much less a few hundred APPLICATIONS.

Leonard Misner: <i>while still processing the entire program, scanning all I/O and updating the image table repeatedly and absolutely reliably at a rate of a few milliseconds</i>

If we're discounting the aforementioned hundreds of applications and talking about one program, this is trivial.

Leonard Misner: <i>even with the plc sitting alone in an oil saturated box for decades at a time.</i>

I think that was supposed to PC, not PLC in that sentence. In other words, a PC wouldn't survive that, but a PLC could. I've never let any of my PLCs get saturated with oil, although some of them have gotten quite dusty thanks to filters never being cleaned. My handful of experiences with water haven't been all that great. "Decades" is probably pushing it. Certainly, 20 years is not unreasonable for a PLC. Much longer than that is an aberration. Most mechanical systems won't last that long anyway. I pulled out a 20 year old PLC-3 and replaced it with a ControlLogix because nobody could work on the damn thing any more, but it was in fact running just fine. I'm not sure what my oldest PC I've replaced is; certainly more than 10 years, but probably less than 20. You may have a point here, but it's still something of an open question. We didn't HAVE ruggedized, fanless, all solid state PCs 20 years ago. Now I do. I have no particular reason to think they won't last as long as a PLC.

Leonard Misner: <i>Test engineers cry whenever I point out that even their test stands (typically pc based controls) can be done easier and more reliably with a plc.</i>

I think this is going to be rather application specific. There are certainly some test applications that a PLC will do great in. However, there are certain things that are just beyond the PLCs reach. LabVIEW can do a lot of really cool things, and handles masses of data quite well. I have trouble envisioning a PLC recording thousands (or millions) of samples per second for days at a time and then analyzing it.

Leonard Misner: <i>One of the biggest reasons the United States has had so many manufacturing plants closed down is because we simply have lost the ability to being competitive in the world market.</i>

Okay. Although that's nearly a tautological statement.

Leonard Misner: <i>You have to get the best quality product out the door at the lowest cost.</i>

Hard to argue with that.

Leonard Misner: <i>It seems so many engineers believe the company they are working for is there to facilitate their own personal technical experimentation and entertainment.</i>

This is not my experience, but I understand your point. We want to ship product, not play with cool toys. However, the original question was not posed in this vein. It was more of a "when all you have is a hammer, every problem looks like a nail." The guy has no experience with PLCs. Apparently no one around him knows anything about PLCs. The PLC is obsolete. Well, he knows he could crank out some C / MatLab / LabVIEW / JavaScript and do this application. I believe that to be a naive solution, but not a self-centered one. My original response was that that was opening a can of worms, and that there are a WHOLE lot of things you have to deal with to take a general purpose device and use it for this particular task. I stand by that. That doesn't mean that everyone who thinks a PC might work is either (a) an idiot, (b) mesmerized by the "shiny object" effect of playing with bleeding-edge tech, or (c) anything other than fully committed to their task.

Leonard Misner: <i>You have mentioned how a PLC is more readily supported, more easily programmed by more people who have that knowledge, ethernet integration is simply a module and copying some config junk from a manual as you suggested.</i>

I'm not sure who the "you" is in this sentence, but it seems to be referring to statements by me. I stand by those statements. I did NOT say "Any one who uses PC-based controls is delusional." Somehow you and Armin both seemed to have gotten that impression.

Leonard Misner: <i>So why would [you] intentionally compromise the future of the company you're working with by forceably integrating a pc with various uniquely concepted techy toys that were NOT DESIGNED SPECIFICALLY TO the task of reliable machine control.</i>

Is this the same "you?" I was essentially making this same argument, so I'm not sure why you're repeating it back to me. I'll answer anyway; I do not believe choosing PC-based control is "intentionally compromising" the future of my company. A customer recently requested a Kuka robot, and we had no problem selling it to them. Kuka uses a PC-based robot controller. They have gone through the work to make their controller a reliable machine control device. It is supported and robust. If a customer asked me to integrate a Beckhoff or B&R system I would have no problem with it. There were reasons my company chose to use PCs for gantry robot control when it started in 1995, and even in hindsight those were valid reasons. (Mainly, PLCs were simply incapable of doing what we needed them to.)

Leonard Misner: <i>The market wants to buy less expensive golden nuggets...not some terd product that was produced by a "black box" that nobody can work with to make that terd product any better than the terd product that it is.</i>

(a) I believe the word you want is "turd."

(b) This line is reminiscent of a scene in Kevin Smith's "Jay and Silent Bob Strike Back." http://www.imdb.com/title/tt0261392/quotes?qt=qt0392693 Not safe for work language. <b>moderator's note: not safe for work language and possibly offensive to some</b>

(c) As you say, who cares what made the product? We want to get product out the door, not sit around arguing about what controller is best. If a PC works, use it. If a PLC works, fine. What matters is product goes out the door.

Leonard Misner: <i>If your stuff can't compete in the market, your company shuts down...welcome to America.</i>

Again, hard to argue. But also not really a point in your favor. What I take that statement to mean is use the best tool for the job. A blind devotion to one solution or another is not in anyone's best interests.

Leonard Misner: <i>Engineers are supposed to be a pretty sharp group... why haven't they figured THAT out?</i>

A rhetorical question, but I'll answer it anyway. Engineers have long been accused of being disconnected from business realities. On the flip side, one could argue that engineers are stretched thin thanks to older works being forced out and a lack of new workers interested in the manufacturing / process sectors.

Leonard Misner: <i>Anybody ever hear of Dr. Deming?</i>

Sure, but I don't think he said anything about PC-based controls. :) Virtually nothing he said impacts this argument, with the possible exception of "don't rely on technology to solve problems." I suppose your argument would be that choosing a PC to save up front costs reflects a lack of long-term planning, or that fiddling with PC tech shows a lack of constancy of purpose. You have a point. As I said originally, grabbing a Dell and copy of Visual Studio is not the right solution. That DOESN'T mean that PLCs are always the right choice, or that a PC is always the wrong one.

-James Ingraham
Sage Automation, Inc.
 
T

Tallak Tveide

> "PC with a simple robust OS" is an oxymoron.
> there is no blue screen of death on a PLC.

I would have to agree on this to a certain degree. But - a stripped down linux installation is a lot simpler than the Microsoft offerings, and I would dare say not much more advanced than the software inside a PLC, with the benefit that the source code is open and therefore may be inspected by anyone who desires it.

> soft PLC, driving animated graphics, handling a hundred analog process
> devices, calculating statistics and creating paretos and productions
> reports...and when you can't even get 4 temperature samples while attempting to

I think we actually agree on this point. A PLC is a quick and simple way to implement many systems, in particular small simple systems. Logging data and controlling a lab process is probably one of these. But the low cost also depends on that the user is knowledgeable on PLC/SCADA/Database setup, and for many people it seems simpler to start with for instance labview and excel on windows, and not have to learn to program a PLC. I would think the results are not as good, but I disagree that a PLC based solution is always cheaper.

What I am talking about is using proper machine separation for 'logic tasks' (in a computer or in a PLC), and for HMI tasks (in a computer or a industrial panel) and also for running the database. As you can see, I would recommend at least three machines in this scenario. So not a lot similarity with the labview+excel setup described above.

And as I also said, while PLC/SCADA based systems are fine, a system based on a programming language could perhaps excel in more complex tasks (dynamic size lists, hashes, sorting, integration with other computer systems etc). But the task of creating such a system is difficult and time consuming, and should not be done for fun in a project, but taken seriously as a project on its own. Then, after the concept is proven, it may be used as a tool in projects that aim to build a plant.
 

> "PC with a simple robust OS" is an oxymoron.
> there is no blue screen of death on a PLC.
[clip]
> A pc is a single tasking slow stupid device which is incapable of having a few hundred process applications running simultaneously while still processing the entire program, scanning all I/O and updating the image table repeatedly and absolutely reliably at a rate of a few milliseconds even with the plc sitting alone in an oil saturated box for decades at a time. <

Never heard about multi tasking, multi threading on multi core CPUs of a PC ? ...

Regards
Armin Steinhoff
 
[ clip]

> ...welcome to America.

IMHO .. the first PC based control systems was used in America. Have a look to the history of QNX and oil platforms.

Best Regards
Armin Steinhoff
 
J

James Ingraham

Steinhoff: <i>IMHO .. the first PC based control systems was used in America. Have a look to the history of QNX and oil platforms.</i>

This may well be correct, although it's probably impossible to find the "first" example. Although I thought QNX's first deployments were in aerospace, but I could be wrong.

Oh, and QNX is Canadian. :)

-James Ingraham
Sage Automation, Inc.

P.S. Armin, what gives? You went 15 rounds with me and we didn't even disagree that much. Then you let this vitriolic rant go through with an off-hand, non-technical response? Is that an insult or a compliment?
 
L

Lynn August Linse

> So I decided to reverse engineer the whole system control and replace it by a
> Pc. I'm thinking of finite state machine or real time control to do this.

(* Smile *)

Quite a political topic! The race to replace PLC with a PC began what ... 15 years ago? For some reason, PLC still exist and most "PC working like PLC" companies have gone bankrupt. Those who say the PC will displace the PLC are kin to those who say IPv6 will replace IPv4 and lots of other forward thinkers :)

You as an individual can of course create a PC app which reads in a temperature.

The biggest problem with "PC as controller" is the hardware changes so fast that it is nearly impossible to provide the exact 'spares' expected for more than a few years (or months) ... unlike the PLC companies who happily charge an arm-and-a-leg (and half the other leg) for a 15-year old replacement part.

I'm sure there are poster-children out there on this list running 15-year old PC. Not I - my oldest CPU is a P4 dual-core which is perhaps 5 years old and runs my RSLogix stuff - and spends 99% of its time powered off. I'm sure I could NOT find any direct replacement parts for it - unless they are used parts on ebay.

All of my active PC systems (Windows or Linux) are running on hardware from 1 to 3 years old at most.
 
>> Steinhoff: IMHO .. the first PC based control systems was used in
>> America. Have a look to the history of QNX and oil platforms.

> This may well be correct, although it's probably impossible to find
> the "first" example. Although I thought QNX's first deployments
> were in aerospace, but I could be wrong.

> Oh, and QNX is Canadian. :)

Oh ... sorry about that !

> -James Ingraham
> Sage Automation, Inc.
>
> P.S. Armin, what gives? You went 15 rounds with me and we didn't
> even disagree that much. Then you let this vitriolic rant go
> through with an off-hand, non-technical response? Is that an
> insult or a compliment?

Two times no. As an "old fighter" of this mailing list I know when I
have to give up :)

Best Regards
Armin Steinhoff
 
J

James Ingraham

Lynn August Linse: <i>...most "PC working like PLC" companies have gone bankrupt.</i>

Huh? Beckhoff, B&R, ABB, and Kuka are just the ones that spring to mind off the top of my head. True, Think & Do and Steeplechase kinda vanished... but (a) they were pure software plays, and (b) neither actually went bankrupt. Both products are still available from current owner Phoenix Contact, though I'm sure their sales are a fraction of their late-90s / early aughts peak. QNX, VxWorks, Advantech, Kontron, etc. all do just fine in this space. National Instruments is going great guns.

Lynn August Linse: <i>Those who say the PC will displace the PLC are kin to those who say IPv6 will replace IPv4 and lots of other forward thinkers :)</i>

Jiminy Cricket! Are we going to have a go at IP v4 / v6 now? Why don't we debate something less controversial, like how to achieve peace in the Middle East.

To your point, there was an initial enthusiasm for PC-based control that was unwarranted. The theory that PCs would eventually dominate PLCs had some serious flaws. The most obvious in hindsight is that the PLC guys weren't just going to sit there and let their products die. Hence the whole "PAC" nomenclature. Bottom line is that PLCs gained a whole host of new features to compete with PCs. Also, at the very low end (a handful of discrete I/O, say) PLCs could compete on cost despite the theoretical price advantage of a PC. The threat of PCs forcing PLC vendors to implement things like motion control, data structures, open networks (well, sorta), web interfaces, etc. is a good thing. So your point is correct, but misses the forest for the trees. PCs will not replace PLCs. However, PCs have thoroughly established themselves as an important option in industrial controls, and forced PLCs to improve.

Lynn August Linse: <i>The biggest problem with "PC as controller" is the hardware changes so fast that it is nearly impossible to provide the exact 'spares' expected for more than a few years (or months) ... unlike the PLC companies who happily charge an arm-and-a-leg (and half the other leg) for a 15-year old replacement part.</i>

One of the major benefits of PCs is that you don't NEED exact spares. I can take software written 15 years ago and run it on a brand-spanking new machine. I did it last week, in fact. Sometimes you have hardware problems; this is where networks really help, especially Ethernet. I don't need a motion control board or a DAQ board or whatever any more. I connect to all my real-world devices via Ethernet. It's true that eventually Ethernet will go away to join serial ports, floppy drives, and AT keyboards. But we've got a LONG time before that happens. And with virtualization, we can keep that old software going even longer.

One final point. In all this back and forth, nobody has brought up the fact there are virtually no industrial control scenarios that DON'T involve a PC. They are frequently used for HMI / SCADA. They are used for programming the PLCs. They're used for troubleshooting, and reporting, and all sorts of things. Does it help you that your PLC is running if the HMI is dead? PCs are already critical pieces of the infrastructure.

-James Ingraham
Sage Automation, Inc.
 
S
<i>One of the major benefits of PCs is that you don't NEED exact spares. I can take software written 15 years ago and run it on a brand-spanking new machine</i>

An excellent point and one that I was going to point out myself.

<i>Does it help you that your PLC is running if the HMI is dead? PCs are already critical pieces of the infrastructure.</i>

Yes. Yes it does. There’s a PC-based package that I use a lot, which is pretty script oriented, and you can see that the publishers are envisioning customers using it for control; they include PID control capability and so on. And yet I just use it for an MMI. No logic is solved, nor any critical information stored on the PC, nor will its sudden, unexpected, or spectacular failure even cause the process to burp.

Roll up another PC, load the runtime license and app, configure the network connection, (hopefully all this has been done beforehand, actually) and you’re online with the process again just like that with no loss of anything nor a bump in the process! Now I’ll admit that if you’re working with a PC-based solution where you DO trust the reliability of the computer hardware, OS, and software to a level comparable to a PLC, just using the PC would yield a more reliable solution as it eliminates one potential point of failure.

I also take issue with your inclusion of programming terminals as critical pieces of the control architecture. Typically, if your laptop dies, it does so while you're doing something other than making online edits, and even in that case, the regard given to reliability <b>by the PLC designers</b> will generally save the day and continue to maintain the process in control till you replace the PC.
 
>> Lynn August Linse: ...most "PC working like PLC" companies have gone bankrupt.

> Huh? Beckhoff, B&R, ABB, and Kuka are just the ones that spring to mind off the top
> of my head. True, Think& Do and Steeplechase kinda vanished... but (a) they were
> pure software plays, and (b) neither actually went bankrupt. Both products are still
> available from current owner Phoenix Contact, though I'm sure their sales are a fraction
> of their late-90s / early aughts peak. QNX, VxWorks, Advantech, Kontron, etc. all do
> just fine in this space. National Instruments is going great guns.

>> Lynn August Linse: Those who say the PC will displace the PLC are kin to those who
>> say IPv6 will replace IPv4 and lots of other forward thinkers :)

> Jiminy Cricket! Are we going to have a go at IP v4 / v6 now? Why don't we debate
> something less controversial, like how to achieve peace in the Middle East.

> To your point, there was an initial enthusiasm for PC-based control that was
> unwarranted. The theory that PCs would eventually dominate PLCs had some serious
> flaws. The most obvious in hindsight is that the PLC guys weren't just going to sit
> there and let their products die. Hence the whole "PAC" nomenclature. Bottom line is
> that PLCs gained a whole host of new features to compete with PCs. Also, at the very
> low end (a handful of discrete I/O, say) PLCs could compete on cost despite the
> theoretical price advantage of a PC. The threat of PCs forcing PLC vendors to
> implement things like motion control, data structures, open networks (well, sorta), web
> interfaces, etc. is a good thing. So your point is correct, but misses the forest for the
> trees. PCs will not replace PLCs. However, PCs have thoroughly established
> themselves as an important option in industrial controls, and forced PLCs to improve.

>> Lynn August Linse: The biggest problem with "PC as controller" is the hardware
>> changes so fast that it is nearly impossible to provide the exact 'spares' expected for
>> more than a few years (or months) ... unlike the PLC companies who happily charge
> an arm-and-a-leg (and half the other leg) for a 15-year old replacement part.

> One of the major benefits of PCs is that you don't NEED exact spares. I can take
> software written 15 years ago and run it on a brand-spanking new machine. I did it last
> week, in fact. Sometimes you have hardware problems; this is where networks really
> help, especially Ethernet. I don't need a motion control board or a DAQ board or
> whatever any more. I connect to all my real-world devices via Ethernet. It's true that
> eventually Ethernet will go away to join serial ports, floppy drives, and AT keyboards.
> But we've got a LONG time before that happens. And with virtualization, we can keep
> that old software going even longer.

Yes, that's the reality ... I share your view.

> One final point. In all this back and forth, nobody has brought up the fact there are
> virtually no industrial control scenarios that DON'T involve a PC. They are frequently used for HMI/SCADA.

We have customers in that area who do complex calculations on a PC instead of a PLC ... requests and responses are communicated between the PC (SCADA) and PLC via PROFIBUS.

Best Regards
Armin Steinhoff
 
A

Armin Steinhoff

>>So I decided to reverse engineer the whole system control and replace it by a
>>Pc. I'm thinking of finite state machine or real time control to do this.

> (* Smile *)

> Quite a political topic! The race to replace PLC with a PC began
> what ... 15 years ago? For some reason, PLC still exist and most
> "PC working like PLC" companies have gone bankrupt.

... and how many PLC companies have gone bankrupt in the same time
frame?

> Those who say the PC will displace the PLC are kin to those who
> say IPv6 will replace IPv4 and lots of other forward thinkers :)

>You as an individual can of course create a PC app which reads in a
>temperature.

When you are using an automated teller machine from VISA .. are you
aware that this system is controlled PC-based?
When you depart from the international airport Frankfurt [FRA] are you aware that the gasoline pumps for the aero plans are controlled by QNX systems with a fieldbus?

Most baggage sorting systems on airports worldwide are PC based controlled.

However ... there are a lot of other PC based solutions which are
supporting your day by day life.

>The biggest problem with "PC as controller" is the hardware changes
>so fast that it is nearly impossible to provide the exact 'spares'

The normal interfaces of PC based control systems are provided by POSIX based operating systems and not by the low level hardware. There are no problems as long as the new version of the PC hardware is supported by the POSIX operating system. MS-Word e.g. is working unmodified on different PCs ... isn't it ? Same story with other applications for other POSIX operating systems.

Yes, after a run of 15 years it's difficult to buy a PC with an ISA or PCI slot ... but there are similar problems to buy an IO interface board for e.g. an old Siemens S5 system.
I don't see here big differences ...

> expected for more than a few years (or months) ... unlike the PLC
> companies who happily charge an arm-and-a-leg (and half the other
> leg) for a 15-year old replacement part.

Yes, why should I buy expensive spare parts for an old PC system when I'm able to buy a newer, faster, cheaper and compatible one ?

> I'm sure there are poster-children out there on this list running
> 15-year old PC. Not I - my oldest CPU is a P4 dual-core which is
> perhaps 5 years old and runs my RSLogix stuff - and spends 99% of
> its time powered off. I'm sure I could NOT find any direct
> replacement parts for it - unless they are used parts on ebay.

What parts do you want to buy? Most components of a PC are on the motherboard and on the CPU (Soc chip) ...

> All of my active PC systems (Windows or Linux) are running on
> hardware from 1 to 3 years old at most.

Well ... and the reason is: You want to use the newest version of
Windows or Linux. And the reasons are not problems with spare parts,
isn't it? :)

Best Regards
Armin Steinhoff
 
J

James Ingraham

Me: <i>Does it help you that your PLC is running if the HMI is dead?</i>

Steve Myres: <i>Yes. Yes it does... No logic is solved, nor any critical information stored on the PC, nor will its sudden, unexpected, or spectacular failure even cause the process to burp.</i>

This may or may not be helfpul. Most of my machines have a Cycle Start button on the HMI. If you lose the HMI, you can't re-start the machine. Granted, our machines probably start / stop a few times a shift, where as some process may go months or years without a stop. So perhaps you can live without your HMI for a little while.

Steve Myres: <i>Roll up another PC, load the runtime license and app, configure the network connection, (hopefully all this has been done beforehand, actually) and you’'re online with the process again just like that with no loss of anything nor a bump in the process!</i>

Again, it depends on the process. What you just described might take days. Just as one example, if you lose a Siemens license in the middle of the night or on a weekend, tough luck. Many software packages can't just be download off of the supplier site, so you have to pray you have the disks somewhere accessible. If there's some special hardware (like a Profibus card) then you could be in for a long lead time. There aren't too many processes that can be left alone for days at a time with no monitoring by a human.

Me: <i>PCs are already critical pieces of the infrastructure.</i>

Steve Myres: <i>I also take issue with your inclusion of programming terminals as critical pieces of the control architecture. Typically, if your laptop dies, it does so while you're doing something other than making online edits, and even in that case, the regard given to reliability <b>by the PLC designers</b> will generally save the day and continue to maintain the process in control till you replace the PC.</i>

I guess it depends on what you mean by "critical." True, if a PLC is running the process and my programming terminal dies I am not likely to blow up Pennsylvania. In that regard, it's not critical. On the other hand, things always go wrong in waves. The process shuts down for some bizarre reason, so you go grab the maintenance laptop. It's dead. So now you can't get online to figure out what's going on. And you can't just re-install the software on another laptop because that one had the license key. Plus, it was the only laptop left with a serial port.

-James Ingraham
Sage Automation, Inc.
 
S
<i>This may or may not be helpful. Most of my machines have a Cycle Start button on the HMI. If you lose the HMI, you can't re-start the machine. Granted, our machines probably start / stop a few times a shift, where as some process may go months or years without a stop. So perhaps you can live without your HMI for a little while. </i>

For any intervention the operator has to do with the machine on a frequent basis like every cycle, I provide hard pushbuttons. $50 to replace vs. the cost of a touchscreen. Plus, some things have to be hardwired anyway, like two-hand inputs, E-Stops, etc. Stuff on the screen is reserved for information and occasional (say few times per hour max) button pushes.

<i>Again, it depends on the process. What you just described might take days. Just as one example, if you lose a Siemens license in the middle of the night or on a weekend, tough luck. Many software packages can't just be download off of the supplier site, so you have to pray you have the disks somewhere accessible. If there's some special hardware (like a Profibus card) then you could be in for a long lead time. There aren't too many processes that can be left alone for days at a time with no monitoring by a human.</i>

Well, like I said, preferably the customer or I has had the foresight to get this all set up on a replacement machine <b>before</b> the primary broke. If you've allowed yourself to have a vulnerability where an easily foreseeable failure will stop your process for days on end, either you haven't finished optimizing the process or you should be looking for a job. Besides, your hypothetical makes it a WORSE idea to use a PC in a process setting, and the post I was responding to basically said it was OK. You're proving MY point.

<i>The process shuts down for some bizarre reason, so you go grab the maintenance laptop. It's dead. So now you can't get online to figure out what's going on.</i>

Yeah, that sounds about like a normal day! ;-)

<i> And you can't just re-install the software on another laptop because that one had the license key. Plus, it was the only laptop left with a serial port.</i>

But that's a strawman IMO. "I'm capable of planning so badly as to create more points of failure in my system, therefore it's safe to use PC's in critical locations" Have a backup plan. Have a backup plan for your backup plan.

Run your automation software from within VM's, preferably VM's on an external hard drive. (Just unplug your hard drive and plug into the replacement laptop. And even if the hard drive isn't external on the original laptop, take it out and use it externally to the replacement -- but remember, VM's on that drive, not a native install) And the VM's should be backed up besides.

As far as the serial port, make sure you have one, or a solution. Test it beforehand and make sure it works.
 
J

James Ingraham

Steve Myres: <i>your hypothetical makes it a WORSE idea to use a PC in a process setting, and the post I was responding to basically said it was OK. You're proving MY point.</i>

Yeah, my message has been a little muddled, hasn't it? (Not the first time.) The whole thread started with me arguing against PCs, then for them, then apparently against them. This schizophrenic back-and-forth is the result of reality being messy. For the original question, I thought grabbing an off the shelf PC and throwing it a control applications was a bad idea. Then someone came back and said never use PCs under any circumstances, and anyone who does is contributing to the Decline and Fall of America. I think this position is far too extreme, and ignores the fact that PCs are already a critical piece of the controls world. You (Steve, for others reading this and getting lost in my ramblings) basically said, "Meh, the way I use PCs they aren't so critical." My examples were to show that PCs are, in fact, critical, even if a momentary blip in one won't necessarily take down your plant. We ALREADY have to deal with PCs as critical pieces of infrastructure. Like you said, we've got to plan ahead and deal with them. Because they're IMPORTANT. Because we basically couldn't run modern facilities without them. They aren't just for accounting and marketing folks, or even the design engineers who never leave their cubicles to see what's happening on the plant floor. The guys in the trenches NEED their PCs. You said (specifically) that PCs for programming do not count as critical; then you said it's important to make sure you have solutions so that a PC failure doesn't wreck your life. That sounds like "critical" to me.

So to clarify; PCs are here to stay, and must be dealt with. There are applications where PCs are a good fit, and ones where they are a bad fit. There are, in particular, some issues with PCs that must be considered when using them in a controls application, and ignoring those issues is going to create long-term problems. There is no one perfect solution for all scenarios.

-James Ingraham
Sage Automation, Inc.
 
J

James Ingraham

We scared him off, I guess. Is that a good thing or a bad thing?

-James Ingraham
Sage Automation, Inc.
 
L

Lynn August Linse

This same topic comes up a few times a years for the decade+ that control.com has been running. Always the same PRO/CON - not much has changed. :)
 
Top